pH selects for distinct N2O-reducing microbiomes in tropical soil microcosms

Abstract Nitrous oxide (N2O), a greenhouse gas with ozone destruction potential, is mitigated by the microbial reduction to dinitrogen catalyzed by N2O reductase (NosZ). Bacteria with NosZ activity have been studied at circumneutral pH but the microbiology of low pH N2O reduction has remained elusive. Acidic (pH < 5) tropical forest soils were collected in the Luquillo Experimental Forest in Puerto Rico, and microcosms maintained with low (0.02 mM) and high (2 mM) N2O assessed N2O reduction at pH 4.5 and 7.3. All microcosms consumed N2O, with lag times of up to 7 months observed in microcosms with 2 mM N2O. Comparative metagenome analysis revealed that Rhodocyclaceae dominated in circumneutral microcosms under both N2O feeding regimes. At pH 4.5, Peptococcaceae dominated in high-N2O, and Hyphomicrobiaceae in low-N2O microcosms. Seventeen high-quality metagenome-assembled genomes (MAGs) recovered from the N2O-reducing microcosms harbored nos operons, with all eight MAGs derived from acidic microcosms carrying the Clade II type nosZ and lacking nitrite reductase genes (nirS/K). Five of the eight MAGs recovered from pH 4.5 microcosms represent novel taxa indicating an unexplored N2O-reducing diversity exists in acidic tropical soils. A survey of pH 3.5–5.7 soil metagenome datasets revealed that nosZ genes commonly occur, suggesting broad distribution of N2O reduction potential in acidic soils.


Introduction
Nitrous oxide (N 2 O) is a long-lived ozone-depleting greenhouse gas with a global warming potential far exceeding that of the equivalent amount of CO 2 [1,2].The global atmospheric N 2 O concentration has increased from 270 parts per billion (ppb) in 1750 to 331 ppb in 2018 [3].During the 2007 to 2016 time period, the net global atmospheric N 2 O increase was estimated at 4.3 Tg N year −1 [4], indicating that N 2 O sources outpace N 2 O sinks.Major sources of N 2 O include denitrification (NO 3 − /NO 2 − → N 2 O) [5] and chemodenitrification [6], with additional N 2 O released from nitrification (NH 4 + → NO 3 − ) [7] and dissimilatory nitrate reduction to ammonium (NO 3 − /NO 2 − → NH 4 + ) [8].Compared to multiple sources of N 2 O, its consumption catalyzed by N 2 O reductase (NosZ) is the major natural biotic sink.Canonical, complete denitrifiers possess the nos operon and can synthesize NosZ responsible for N 2 O reduction to dinitrogen (N 2 ), the latter a gas without warming potential.Genomic analyses distinguished two types of nos operons with distinct nosZ: Clade I nosZ generally associated with canonical denitrifying bacteria and Clade II nosZ often found on genomes lacking the denitrification biomarker genes nirS or nirK [9][10][11].Subsequent studies reported that Clade II nosZ are generally more abundant and diverse in soils than Clade I nosZ [12,13], suggesting N 2 O reduction potential outside the group of complete denitrifiers.Kinetic studies using axenic bacterial cultures demonstrated that Clade II N 2 O reducers exhibit higher affinities to N 2 O and growth yields than their N 2 O-reducing Clade I counterparts [14], suggesting Clade II reducers capture energy during growth with N 2 O as electron acceptor more efficiently.Environmental factors, including pH, temperature, O 2 levels, substrate availability, and NO 3 − /NO 2 − levels are known to affect N 2 O reduction [15][16][17].The final reduction step (N 2 O → N 2 ) catalyzed by NosZ was found particularly sensitive to pH, explaining N 2 O emissions from acidic environments [18,19].Expression studies suggested posttranscriptional interference at low pH, with the NosZ remaining in the catalytically inactive apoform, a possible reason for the observed decline in NosZ activity under acidic conditions [18,20].Natural soils are the dominant source of N 2 O (∼5.6 Tg N 2 O-N year −1 ), accounting for ∼33% of global N 2 O emissions [4], with tropical forest soils contributing ∼1.34 Tg N 2 O-N year −1 [21].Tropical soils were reported to emit N 2 O at a rate of 0.1 ± 0.04 g N 2 O-N m −2 year −1 , which is about 50% above the average rate of global soil N 2 O emissions [22].The large contribution of tropical soils to N 2 O emissions has been explained by high N 2 fixation activity and generally acidic soil pH [23][24][25].N 2 O production can be sporadic suggesting that f luxes and concentrations of N 2 O can vary substantially both over temporal and spatial scales [26].A comparative metagenomic study found similar relative abundances of nosZ sequences in acidic tropical and circumneutral temperate soils [23].Apparently, acidic tropical soil environments have the metabolic potential to reduce N 2 O, a hypothesis supported by isotopic measurements that revealed biotic reduction of N 2 O in acidic tropical forest soils [24].Although some evidence for N 2 O reduction under acidic conditions exists, the consensus backed by laboratory observations is that N 2 O reduction is negligible at acidic pH [18,20,27,28].
To reconcile existing inconsistencies between field measurements and laboratory studies, and to explore the impacts of pH and N 2 O concentration on N 2 O-reducing microorganisms, acidic (pH 4.4-5.0)soil samples were collected in the Luquillo Experimental Forest (LEF) in Puerto Rico.A series of microcosms explored the impact of pH (i.e.

Soil collection
Soil samples were collected from four locations (5-20 cm depth) in the LEF in Puerto Rico, including the El Verde tabonuco forest (EV, 453 m above mean sea level [MSL]), the Palm Nido palm forest (PN, 634 m MSL), the Pico del Este elfin forest (PE, 953 m MSL), and the Sabana tabonuco forest (S, 265 m MSL) (Supplementary Fig. S1).The soil materials were transferred to sterile Whirl-Pak bags, placed at 4 • C, and manually homogenized prior to microcosm setup.Detailed descriptions of the LEF can be found elsewhere [29,30], and physicochemical properties of the soil samples are presented in Supplementary Table S1.

Soil microcosms
Completely synthetic, reduced (0.2 mM l-cysteine) mineral salt medium was prepared following established protocols [31].In the pH 4.5 medium, 50 mM potassium dihydrogen phosphate replaced the 30 mM bicarbonate buffer used in the pH 7.3 medium [32].The pH was adjusted with CO 2 (pH 7.3 medium) or with 4 M hydrochloric acid (pH 4.5 medium).CuCl 2 (17 μM) and the Wolin vitamin mix were added from concentrated stock solutions to individual 160 mL glass serum bottles after autoclaving [32].
Inside a glove box (Coy Laboratory Products, Grass Lake, MI) filled with 97% N 2 and 3% H 2 , 2 g (wet weight) of homogenized soil material was aseptically transferred to sterile glass serum bottles containing 100 ml of medium using stainless-steel spatulas.The serum bottles were immediately resealed with sterile butyl rubber stoppers, crimped with aluminum caps, and removed from the glove box.Lactate (5 mM) was added to each serum bottle from a 1 M stock solution by syringe and replenished four times over the 15-month incubation period.Two replicate series of microcosms were established at pH 4.5 and at pH 7.3 with duplicate pH 4.5 microcosms (Supplementary Fig. S1).Plastic syringes (Becton, Dickinson and Company, Franklin Lakes, NJ) with 25gauge needles (Becton, Dickinson and Company) were used to add 0.1 ml (4.17 μmol, 0.02 mM aqueous N 2 O) and 10 ml (416.7 μmol, 2 mM aqueous N 2 O) of undiluted N 2 O gas to the incubation vessels.N 2 O was periodically analyzed by gas chromatography (GC) as described [32] and replenished when consumed.All bottles were incubated at 30

Analytical procedures
N 2 O was analyzed with an Agilent 3000A Micro-GC (Agilent, CA) equipped with a thermal conductivity detector and a Plot Q column [14].The limit of detection was 50 ppmv of N 2 O with signal-to-noise ratio of 3:1.The injector and column temperatures were set to 100 • C and 50 • C, respectively, and the column pressure was set to 25 psi.For each measurement, a 0.1 ml headspace sample was withdrawn from the microcosm and manually injected into the Micro-GC.Aqueous N 2 O concentrations were calculated from the headspace concentration using a dimensionless Henry's constant for N 2 O at 20 • C of 1.68 based on the equation C aq = C g /H cc [33].C aq and C g are the aqueous N 2 O and the headspace N 2 O concentrations (μM), respectively, and H cc is the dimensionless Henry's constant.The total amount of N 2 O was calculated as the sum of N 2 O in the headspace and the aqueous phase.

DNA extraction and metagenome sequencing
When about half of the final N 2 O amendment had been consumed, the microcosms were shaken, and 5 ml suspension samples were collected with 5-ml plastic syringes equipped with 18gauge needles.DNA for shotgun metagenome sequencing was extracted with the DNeasy PowerSoil kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions.DNA concentrations were determined using the Qubit f luorometer (Life Technologies, Carlsbad, CA).Metagenome sequencing was performed at the Institute for Genome Sciences at the University of Maryland using the Novaseq 6000 platform (Illumina, San Diego, CA) to generate 48 to 73 million reads with 150-bp read length per sample (Supplementary Table S2).Replicate high-and low-N 2 O pH 4.5 microcosms showed similar N 2 O reduction performance, and a single microcosm per treatment was randomly selected for metagenome sequencing.

Bioinformatic analysis
The metagenomes of the four original soils had been sequenced previously [23], and were downloaded from European Nucleotide Archive under project PRJEB26500.The raw reads of the original soils and of the 16 N 2 O-reducing microcosms were trimmed with Trimmomatic v0.39 using default parameters [34].Subsequent assembly was performed using IDBA-UD v1.1.3[35], and only contigs longer than 1000 bp were included in downstream analyses.Contigs were binned using MaxBin2 v2.2.4 with default settings to recover individual metagenome-assembled genomes (MAGs) [36].MAGs were dereplicated with dRep using default parameters [37] and checked for completeness and contamination using CheckM v1.0.18 [38].The resulting MAGs were evaluated for their intrapopulation diversity and sequence discreteness using fragment recruitment analysis scripts available through the Enveomics collection [39].The coverage of each MAG in each metagenome was calculated by estimating sequencing depth at each position using Bowtie 2 [40] with default settings for read mapping.In addition, BEDTools [41] was used to calculate the average of the central 80% of the distribution, which removes the highest and lowest 10% of outlier positions in terms of coverage (i.e. the truncated average depth [TAD80]).TAD80 is a conservative metric that produces no false-positive results [42].TAD80 values were normalized by the genome equivalent of the corresponding metagenome to determine the relative abundance.The genome equivalents of each metagenomic dataset were obtained using the MicrobeCensus package [43].

Metagenomic community profiling
GraftM v0.13.1 was used to extract 16S rRNA gene fragments from the trimmed metagenomic datasets for classification using the Greengenes database (release 13_8) at the 97% nucleotide identity level [44,45].The relative abundance of operational taxonomic units (OTUs) was calculated based on the number of reads assigned to each OTU.Community profiling was based on OTU taxonomic assignments at the phylum, family, and genus levels.

Identification of nosZ genes
ROCker was used to identify metagenomic reads carrying nosZ [46].Brief ly, trimmed short reads were used as the query for BLASTX (Diamond v0.9.14.115) searches against the corresponding ROCker protein database representing the target gene [47].The matching sequences were then filtered using the ROCker compiled models available through the Enveomics collection (http:// enve-omics.ce.gatech.edu/)[39].The abundances of target genes (i.e.Clade I and Clade II nosZ) were determined by calculating the ratio between normalized target reads (counts divided by the median protein length) and the genome equivalents [48].Reference nosZ genes were also searched against the assemblies and MAGs using precompiled hidden Markov models obtained from FunGene and HMMer [49,50].Hits with an identity value of 100% were filtered based on the NosZ sequences in the reference database [23].

NosZ phylogeny
NosZ reference sequences were aligned with ClustaloΩ using default settings [51].The alignment was used to build a maximum likelihood reference tree in RAxML V8.2.12 with "-f a" algorithm, gamma parameter optimization, and a general time reversible model option [52].ROCker identified reads carrying nosZ, which were translated to protein sequences using MetaGeneMark [53].The translated sequences were added to the NosZ reference protein alignment using MAFFT and the "addfragments" option [54], and the new alignment placed in the NosZ reference phylogenetic tree using RAxML EPA algorithm (−f v option).The generated jplace file was processed using an in-house script (available through http://enve-omics.ce.gatech.edu/)for visualization in iTOL [55].Detected NosZ proteins in the assemblies and MAGs were manually curated, added to the phylogenetic trees, and visualized as described above.

N 2 O reduction potential in acidic soils
From MAGs and contigs from acidic N 2 O-reducing microcosms, 33 near full-length NosZ sequences were recovered (Supplementary Figs S2 and S3), and dereplication yielded eight unique NosZ sequences.A customized database was generated that comprised 14 NosZ, including six closely related sequences from the reference NosZ database.To identify nosZ genes recovered from acidic N 2 O-reducing microcosms in various soil microbiomes, 35 available soil metagenome datasets derived from representative pH 3.5-7.5 environmental systems were downloaded from the European Nucleotide Archive (Supplementary Table S3) and subjected to BLASTX query using the customized NosZ database with a minimum identity value of 60% and an e-value of 1e-05.The normalization approach described above was used to calculate relative abundances based on nosZ-carrying metagenomic reads.

Statistics
Statistical analyses were performed using R version 4.0.2.Beta diversity was calculated using Bray-Curtis dissimilarity and visualized using the principal coordinate analysis (PCoA) plot in R with packages ggplot2 [63] and phyloseq [64].Statistical differences in microbial communities among original soils, pH 4.5 microcosms, and pH 7.3 microcosms were determined using permutational multivariate analysis of variance (PERMANOVA) using the Adonis function in vegan with 999 permutations [65].Heatmaps of annotation results of MAGs were generated using the R package pheatmap.

N 2 O reduction in tropical forest soil microcosms
N 2 O consumption occurred with lag times of 1-2 weeks in pH 4.5 and pH 7.3 microcosms with low level of 0.02 mM N 2 O. Longer lag times of up to 7 months were observed in microcosms with 2 mM N 2 O under both pH conditions, but microcosms with high-level N 2 O consumed substantially more N 2 O over the 15-month incubation period (Supplementary Fig. S4 and Supplementary Table S4).No N 2 O loss occurred in autoclaved microcosms, indicating the acidic tropical forest soils harbor N 2 O-reducing microorganisms.In addition, no N 2 O was detected in control microcosms without N 2 O amendment, indicating that N 2 O formation from nitrogenous compounds in the medium or associated with the soil did not occur or was negligible.

Bacterial community composition
Following a 15-month incubation period and when the microcosms had consumed about half of the final N 2 O feeding, DNA was extracted for metagenome sequencing.The overarching goal was to assess if the enrichment conditions selected for different N 2 O-reducing taxa, rather than a fine-scale analysis of the overall community responses over the 15-month incubation.Therefore, single microcosms per treatment were subjected to metagenome sequencing.Totals of 5963 ± 2503 and 23 399 ± 15 030 sequences representing 16S rRNA genes were obtained from the original soil and the corresponding microcosm metagenome datasets, respectively, and yielded 903 16S rRNA gene-based OTUs (Supplementary Table S5).Rarefaction analysis suggested that the number of unique 16S rRNA genes approached saturation in most samples (Supplementary Fig. S5), and the number of OTUs detected in N 2 O-reducing microcosms was lower than in the corresponding original soils.Beta diversity analysis using the 16 microcosm-and four original soil-[23] derived metagenomic datasets indicated distinct community compositions in response to pH and N 2 O levels (P < .01,PERMANOVA) (Fig. 1A, Supplementary Table S6).PCoA revealed that ∼55% of the total variability of OTUs observed in pH 4.5 and in circumneutral N 2 O-reducing microcosms compared to the respective original soils was explained by pH and N 2 O (Fig. 1A).Datasets acquired from same pH microcosms established with the four different soils clustered together indicating that pH shaped distinct microbial communities over the 15-month incubation period (Fig. 1).The majority of the 16S rRNA gene sequences derived from the original soils affiliated with the phyla Proteobacteria, Actinobacteria, Verrucomicrobia, Planctomycetes, and Acidobacteria, with a combined relative abundance exceeding 70% (Fig. 1B).Higher relative abundances of Proteobacteria, Firmicutes, Chlorof lexi, and Actinobacteria were observed in the 16 N 2 O-reducing microcosms relative to the original soil inoculum.Analysis at the family level showed that sequences representing Peptococcaceae, Veillonellaceae, Clostridiaceae, and Hyphomicrobiaceae increased in acidic microcosms, and sequences of Rhodocyclaceae, Clostridiaceae, Hyphomicrobiaceae, and Ruminococcaceae were more abundant in microcosms maintained at circumneutral pH compared to the original soils (Supplementary Fig. S6A).Based on the relative abundances of 16S rRNA gene sequences (highest observed values shown in parentheses), the genera Desulfosporosinus (45%), Desulfomonile (5%), Rhodoplanes (53%), Azospira (56%), and Dechloromonas (38%) increased in response to N 2 O additions compared to the original soils (Supplementary Fig. S6B).At least some members of these genera comprise known N 2 O-respiring species [9], suggesting that bacteria capable of using N 2 O as an electron acceptor were enriched.(Supplementary Fig. S7, Supplementary Table S7).Placing Clade II nosZ sequence reads extracted from the original soil metagenome datasets in the reference Clade II nosZ phylogenetic tree revealed affiliations with the genera Anaeromyxobacter and Opitutus (Fig. 2A).In contrast, the nosZ sequence reads observed in the acidic microcosms were assigned to the genera Profundibacter, Desulfosporosinus, and Desulfomonile (Fig. 2B and C).In microcosms maintained at pH 7.3, the majority of nosZ reads affiliated with the genera Azospira, Dechloromonas, and Sulfuricella (Fig. 2D and E), except for the EV soil microcosms with high level of N 2 O, where nosZ sequences assigned to the genus Desulfosporosinus dominated.A comparative analysis of Clade II nosZ reads revealed that the microcosms maintained at pH 4.5 and at pH 7.3 developed distinct N 2 O-reducing communities (P < .01,PERMANOVA) (Supplementary Fig. S8A).Taken together, these analyses suggest that bacteria with Clade II nosZ drive N 2 O reduction in all tropical soil microcosms, and pH selects for distinct Clade II N 2 O reducers (Fig. 2).

Phylogenetic distribution and relative abundance of Clade II versus Clade I nosZ
Most of the Clade I nosZ sequences could be assigned to the genus Bradyrhizobium, and a small number of reads affiliated with the genera Methylocystis, Methylocella, and Janthinobacterium, independent of the pH conditions or N 2 O levels (Fig. 2F-J).Apparently, acidic versus circumneutral pH and low versus high N 2 O levels did not select for distinct Clade I nosZ N 2 O reducers.The PCoA supports that acidic pH and circumneutral pH conditions selected for similar Clade I N 2 O reducers (P > .05,PERMANOVA) (Supplementary Fig. S8B).

Metagenome-assembled genomes
A non-redundant set of 17 high-quality MAGs (>75% completeness, <5% contamination) harboring nosZ genes was recovered from the 16 metagenome datasets generated from the N 2 O-reducing microcosms (Fig. 3A, Supplementary Tables S8  and S9).Eight of these 17 MAGs have high relative abundance (TAD80 > 1%) in the corresponding metagenomes, revealing that these MAGs represented abundant members of the enriched communities (Fig. 3B).Only five MAGs with nosZ were detected in the metagenomic data of the original soils with relative abundance of ranging between 0.01% and 0.02% (Supplementary Fig. S9), much lower than the relative abundances observed in the corresponding microcosms (Fig. 3B).The relative abundance of MAGs derived from acidic microcosms in the corresponding (i.e.same soil sample) circumneutral microcosms was negligible, as was the relative abundance of MAGs derived from circumneutral microcosms in the corresponding acidic microcosms (Supplementary Fig. S10).This observation indicates the strong impact of pH on the enrichment of N 2 O-reducers in microcosms that received N 2 O.
The AAI results assigned all 17 MAGs harboring nosZ genes to novel taxa (Supplementary Table S10), indicating an unexplored diversity of N 2 O reducers exists in the tropical forest soils and can be enriched in the N 2 O-reducing microcosms.Six of eight MAGs recovered from acidic N 2 O-reducing microcosms were assigned to the genera Desulfitobacterium, Rhodoplanes, and Desulfosporosinus with AAI values exceeding 68% (Supplementary Table S10), probably representing novel species [66].The AAI similarities between the remaining two MAGs, PE_pH4.5_10_MAG7 and MAG8, and the corresponding closest relatives are 45.04% and 44.04%, respectively, suggesting these MAGs represent novel families (Fig. 3A and Supplementary Table S10).MAGs obtained from circumneutral, N 2 O-reducing microcosms were taxonomically related to the genera Azospira, Sulfurimicrobium, Dechloromonas, and Desulfitobacterium (Supplementary Table S10).

Functional analysis of MAGs harboring nosZ genes
Key metabolic pathways or functions of the 17 MAGs harboring nosZ were predicted based on KEGG annotation (Fig. 3C).All eight MAGs obtained from acidic N 2 O-reducing microcosms lack the hallmark denitrification genes nirK/nirS, which encode nitrite reductase.This indicates that these organisms are non-denitrifiers.In contrast, seven of the nine MAGs derived from circumneutral microcosms carried a complete set of genes, including nirS, for canonical denitrification (i.e.NO 3 . In addition to nosZ, other genes of the nos cluster were detected in all MAGs (Supplementary Fig. S11).
These findings indicate that non-denitrifying N 2 O-reducing bacteria may be the main drivers for N 2 O reduction in low pH soils.NosZ is a copper-containing enzyme and the extracellular copper concentration controls nosZ expression [67].Notably, genes for copper transport were identified in 14 of the 17 MAGs harboring nosZ.Sources of electrons for reductive processes, such as N 2 O reduction, were exogenously added lactate and organic material associated with the soil.Lactate was readily consumed in all live microcosms.The analysis of PE_pH4.5_10_MAG6 and MAG7 recovered from acidic N 2 O-reducing microcosms revealed genes encoding l-lactate dehydrogenase (LDH, K00016), which are implicated in the conversion of lactate to pyruvate.Only PE_pH4.5_10_MAG6contained the complete set of genes (pta, K13788, and acyP, K01512) required to metabolize lactate to acetate (Fig. 3C).In addition, five MAGs, three from acidic and two from circumneutral pH microcosms, possessed genes implicated in the fermentation of pyruvate to formate and acetyl-CoA (Fig. 3C).These findings suggest that most MAGs harboring nosZ genes were unable to directly utilize lactate as an electron donor.
Collectively, these results demonstrate that MAGs harboring nosZ genes are taxonomically and metabolically diverse.

nosZ genes in metagenomes representing low pH soil biomes
A total of 27 metagenomes from low pH (pH 3.5-5.7)forest, agricultural, and permafrost soil ecosystems were analyzed (Supplementary Table S3), and all of them harbored nosZ sequences.nosZ abundances ranged from 0.002 ± 0.001 to 0.25 ± 0.11 genome equivalents (i.e. the fraction of genomes expected to carry nosZ assuming one gene copy per genome; Fig. 4), suggesting that the respective acidic soils have N 2 O reduction potential.The highest abundances of nosZ genes were observed in the permafrost soil, and the lowest abundances were detected in two temperate forest soils.nosZ genes were also abundant in low pH tropical forest soils, which typically have high N 2 fixation and nitrogen turnover activities.An expanded survey that included the 27 metagenomic datasets from low pH plus eight datasets from pH 6 to 7.5 soils revealed that nosZ sequences representing low pH soil biomes were predominantly found in pH 4.5-6 environments (Supplementary Fig. S12).The number of metagenome datasets from acidic soils is currently limited; however, the observation that nosZ genes occur in all acidic soils included in the analysis suggests broad distribution of low pH microbial N 2 O reduction potential.requires the formation of new biomass (i.e.growth) under the prevailing environmental conditions.Laboratory studies with consortia that utilize the toxin vinyl chloride as respiratory electron acceptor illustrate this issue.Reductive dechlorination activity was observed at pH <5.5 [73]; however, this activity relied on existing biomass produced at circumneutral pH, and growth of vinyl chloride-respiring Dehalococcoides mccartyi strains did not occur at acidic pH, indicating that a sustainable process in acidic groundwater cannot be envisioned [73].In analogy, it is uncertain if the results of short-term microcosm studies without repeated N 2 O feedings generate meaningful information to predict in situ N 2 O reduction activity in low pH soils.Further complicating data interpretation is the observation that common oxygen-respiring bacteria (e.g.members of the Gemmatimonadaceae) utilize N 2 O as an electron sink following oxygen depletion; however, this process is uncoupled from growth and not sustained under anoxia [75,76].The LEF soil microcosms were incubated for 15 months, and repeated N 2 O additions were consumed at accelerating rates, an observation inconsistent with the activity of residual (i.e.nongrowing) biomass and indicative of respiratory N 2 O utilization and growth of N 2 O reducers at pH 4.5.Our observations challenge the notion that efficient N 2 O reduction requires circumneutral pH and suggest that N 2 O reduction can be sustained under acidic pH conditions.has been demonstrated [6]; however, microcosms maintained at circumneutral pH selected for different N  S10) are responsible for N 2 O reduction (Fig. 3 and Supplementary Fig. S3).Prior studies reported that N 2 O reduction under slightly acidic conditions (pH ∼6.0) was driven by complete denitrifiers that represent cultivated taxa [83,84].Consistent with these prior reports, the MAGs with nosZ derived from microcosms maintained at circumneutral pH represent complete denitrifiers.Of note, three MAGs derived from acidic microcosms with low level of N 2 O each harbored two nosZ genes, one Clade I and one Clade II, a genotype observed for few species of the class Betaproteobacteria capable of denitrification at circumneutral pH [85].Collectively, the findings show that circumneutral pH selects for complete denitrifiers, whereas acidic pH selects for  S3; global soil pH data were obtained from the Soil Geographic Databases (https://www.isric.org); the plots show the total abundance of nosZ genes per genome equivalent; the black lines point to the approximate locations from where the soil metagenomes were derived.

Impact of enrichment on N 2 O-reducing taxa
non-denitrifying N 2 O reducers in the tropical forest soils studied here.A relevant conclusion from this observation is that complete denitrifiers may not represent representative models to study low pH N 2 O reduction.

Impact of N 2 O concentration on N 2 O reduction
A striking difference in microcosms with 0.02 mM versus 2 mM N 2 O was the extended lag phase observed in the high-level N 2 O microcosms independent of the pH condition.A possible explanation for the delayed start of N 2 O consumption is the inhibition of corrinoid-dependent pathways.Micromolar concentrations of N 2 O were shown to repress methionine biosynthesis [67], methanogenesis [86], methylmercury formation [87], and bacterial reductive dechlorination [32], all processes that involve enzymatic steps that strictly depend on the cobalt (I) supernucleophile, a species highly susceptible to oxidation by N 2 O [88].An initial N 2 O concentration of 2 mM exceeds the reported inhibitory constants for corrinoid-dependent enzymes about 100-fold, suggesting the disruption of metabolic pathways directly or indirectly impacted N 2 O-consuming populations.Only a subset of bacteria and archaea synthesize corrinoids [89], and corrinoid prototrophs supply this essential nutrient to corrinoid auxotrophs [90,91].The addition of N 2 O disrupts these microbe-microbe interactions, which are inf luenced by the concentration of N 2 O. Also possible is that the delayed onset of N 2 O reduction in microcosms with high level of N 2 O ref lects a switch to corrinoid-independent metabolic pathways [32,67].

Implications for N 2 O reduction in low pH environments
Global atmospheric N 2 O concentrations are on the rise, and the acidification of agricultural and forest soils, two major sources of atmospheric N 2 O, is predicted to exacerbate emissions [4,92,93].Soils with a pH below 5.5 currently comprise ∼30% of the global ice-free land area and are mainly distributed in the northern temperate to cold belt and the southern tropical belt [94].Our findings suggest that yet-to-be characterized, non-denitrifying bacterial taxa catalyze N 2 O reduction at acidic pH.The microbial N 2 O reduction potential appears to be distributed in acidic soils (Fig. 4), a finding likely to be substantiated as more metagenomes representing low pH microbiomes become available.These observations seem at odds with the dogma that acidic environmental systems are predominantly N 4.5 versus circumneutral) and N 2 O concentrations (i.e.0.02 and 2 mM) on N 2 O reduction.Metagenome analyses indicate that both pH and N 2 O concentration select for distinct N 2 O-reducing microbiomes and suggest widespread distribution of N 2 O reduction potential across various acidic soil ecosystems.

N 2 O
reduction is catalyzed by NosZ and nosZ gene abundances increased during enrichment with N 2 O.The analysis of the metagenome datasets showed that Clade II nosZ sequences outnumbered Clade I nosZ gene sequences in the original soils and in the N 2 O-reducing microcosms (except microcosm S_pH7.3_0.1)

Figure 1 .
Figure 1.Microbial community composition of the original tropical forest soils and the N 2 O-reducing microcosms maintained under low versus high levels of N 2 O and at pH 4.5 versus pH 7.3; (A) beta diversity of microbial communities based on weighted Unifrac analysis of 16S rRNA gene fragments recovered from the metagenomes; samples are visualized by PCoA with colors distinguishing original soils and microcosms at acidic versus circumneutral pH; the ellipses represent the 95% confidence intervals; (B) the relative abundance distributions of the top 20 phyla observed in the original soils and the different microcosms; the numerals shown on the x-axis labels indicate the volumes of N 2 O added with each feeding to the microcosms (0.1 ml [0.02 mM aqueous N 2 O] versus 10 ml [2 mM aqueous N 2 O]).

Figure 2 .
Figure 2. Phylogenetic diversity of nosZ reads recovered from the four original soils and the 16 microcosms maintained at pH 4.5 versus pH 7.3 and at low versus high levels of N 2 O; trimmed Clade II and Clade I nosZ reads in the metagenomes were identified by ROCker and placed in the corresponding reference nosZ phylogenetic tree, as described in the Methods section; the radii of the pie charts are proportional to the number of reads assigned to each nosZ subclade and the colors represent the different soils; (A) diversity of Clade II nosZ reads from metagenomes of the four original soil samples; (B) phylogenetic information of Clade II nosZ in N 2 O-reducing microcosms at acidic pH with low level of N 2 O; phylogenetic information of Clade II nosZ in microcosms at acidic pH with high level of N 2 O (C), at circumneutral pH with low level N 2 O (D), and at circumneutral pH with high level N 2 O (E); (F) diversity of Clade I nosZ reads from metagenomes of the four original soil samples; (G) phylogenetic information of Clade I nosZ in N 2 O-reducing microcosms at acidic pH with low level of N 2 O; phylogenetic information of Clade I nosZ in microcosms at acidic pH with high level of N 2 O (H), at circumneutral pH with low level N 2 O (I), and at circumneutral pH with high level N 2 O (J).

Figure 3 .
Figure 3. Phylogenetic and functional analysis of MAGs harboring nosZ genes; panel (A) shows a phylogenetic tree of the 17 MAGs harboring a nosZ gene and their closest neighbors based on the analysis of 120 bacterial marker genes; MAGs recovered from pH 4.5 and pH 7.3 N 2 O-reducing microcosms are shown in red versus green font, respectively; panel (B) shows the relative abundance of each MAG harboring a nosZ gene in the corresponding N 2 O-reducing microcosms; the relative abundance was calculated by normalizing TAD80 values to genome equivalents; the abundance of these MAGs in the original soil metagenomes was below 0.01% panel (C) depicts a heatmap showing the completeness of key metabolic pathways or functions in the 17 MAGs harboring a nosZ gene based on KEGG annotation; the numbers 0.1 and 10 in the x-axis labels represent the low and high levels of N 2 O added to the microcosms, respectively.

Figure 4 .
Figure 4. Distribution of N 2 O reduction potential in acidic (pH 3.5-5.7)soils; detailed information about the metagenome datasets is provided in Supplementary TableS3; global soil pH data were obtained from the Soil Geographic Databases (https://www.isric.org); the plots show the total abundance of nosZ genes per genome equivalent; the black lines point to the approximate locations from where the soil metagenomes were derived.
• C under static conditions for 15 months.Negative controls included heat-killed (autoclaved) replicates and microcosms without N 2 O but with lactate at pH 4.5 for each soil sample.

pH paradigm for microbial N 2 O reduction
73,74]uction to occur in such micro-environments [18,73,74].Further, the interpretation of observations made in shortterm soil microcosm incubations is not straightforward because it is unclear if the N 2 O reduction activity observed under acidic pH conditions is due to residual activity of existing biomass, or linked to the formation of new cells that conserve energy for growth from N 2 O reduction.The sustainable reduction of N 2 O Prior metagenomic work demonstrated that soils collected at different locations (i.e.elevations) in the LEF share the predominant Clade II nosZ genes and implicated similar taxa in N 2 O reduction[23].Enrichment during the 15-month microcosm incubation revealed that different Clade II type N 2 O reducers are responsible for N 2 O reduction in the LEF soil samples collected at different elevations (Figs2 and 3).For instance, the metagenome analysis of the original soils prior to enrichment implicated Anaeromyxobacter populations as predominant N 2 O reducers (Fig.2).N 2 O reduction by Anaeromyxobacter at circumneutral pH

Non-denitrifying N 2 O reducers responsible for N 2 O reduction under low pH conditions
2 O sources, and future research should address two major knowledge gaps.Although N 2 O pools (i.e.how much is there?) can be quantitatively captured, robust tools to measure N 2 O f luxes (i.e. the rates of N 2 O formation and consumption) that ultimately determine emissions are lacking.Thus, it is not obvious if N 2 O emissions from acidic soil ecosystems ref lect a lack of microbial N 2 O consumption capacity or an imbalance between N 2 O formation versus consumption.Very limited information is available about the taxonomic diversity, physiology, and ecology of microbes that reduce N 2 O under acidic conditions.This study identified several genera (i.e.Desulfosporosinus, Desulfitobacterium, Desulfomonile, and Rhodoplanes) as potential targets for more detailed exploration of low pH N 2 O reduction, and future efforts should focus on the isolation and characterization of N 2 O reducers from low pH environments to generate detailed process understanding.wanella loihica strain PV-4.ISME J 2015;9:1093-104.https://doi.org/10.1038/ismej.2014.20118. Liu B, Frostegård Å, Bakken LR.Impaired reduction of N 2 O to N 2 in acid soils is due to a posttranscriptional interference with the expression of nosZ.mBio 2014;5:e01383-14.https://doi.org/10.1128/mBio.01383-1419.Van Den Heuvel R, Van Der Biezen E, Jetten M et al.Denitrification at pH 4 by a soil-derived Rhodanobacter-dominated community.Environ Microbiol 2010;12:3264-71.https://doi.org/10.1111/j.1462-2920.2010.02301.x20.Frostegård Å, Vick SH, Lim NY et al.Linking meta-omics to the kinetics of denitrification intermediates reveals pH-dependent causes of N 2 O emissions and nitrite accumulation in soil.ISME J 2022;16:26-37.https://doi.org/10.1038/s41396-021-01045-221.Werner C, Butterbach-Bahl K, Haas E et al.A global inventory of N 2 O emissions from tropical rainforest soils using a detailed biogeochemical model.Glob Biogeochem Cycles 2007;21:21.https:// doi.org/10.1029/2006GB00290922. Tian H, Yang J, Xu R et al.Global soil nitrous oxide emissions since the preindustrial era estimated by an ensemble of terrestrial biosphere models: magnitude, attribution, and uncertainty.Glob Chang Biol 2019;25:640-59.https://doi.org/10.1111/gcb.14514 23.Karthikeyan S, Orellana LH, Johnston ER et al.Metagenomic characterization of soil microbial communities in the Luquillo experimental forest (Puerto Rico) and implications for nitrogen